INDEX

Explanations

occurrences of the word "lock"

oai_token-act-pair · gpt-3.5-turbo

references to locks and locking mechanisms

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Eth

-0.81

 Paula

-0.76

 Trop

-0.76

bia

-0.75

bert

-0.74

aukee

-0.71

å¹

-0.70

athy

-0.68

 Crawford

-0.68

Beg

-0.68

POSITIVE LOGITS

 lock

3.70

 locks

2.80

 locking

2.28

Lock

2.26

 Lock

2.26

lock

2.15

 locked

2.00

locks

1.83

 lockdown

1.80

LOCK

1.64

Activations Density 0.006%

occurrences of the word "lock"

references to locks and locking mechanisms

No Comments

No Known Activations